22 juin
22/06/2017 14:00


Soutenance de thèse : Ge XIN

Sparse Representations in Vibration-based Rolling Element Bearing Diagnostics

Doctorant : Ge XIN 

Laboratoire INSA : Laboratoire Vibrations Acoustique (LVA)
Ecole doctorale :  EDA162 : Mécanique, Energétique, Génie civil et Acoustique (MEGA)

Although vibration-based rolling element bearing diagnostics is a very well-developed field, the research on sparse representations of vibration signals is yet new and challenging for machine diagnosis. As a desired property -- representation of a signal in highly organized structure - sparsity enables us to reveal the natural signature of singular events embedded in a signal so as to reduce the demand on the user's expertise, even though it involves advanced theory of stochastic processes. In this thesis, several novel methods have been developed so as to serve the industry in rolling element bearing diagnostics.
First, the sparsity-based model is investigated based on the current literature. An interpretation of sparse structure in the Bayesian viewpoint is proposed for machinery fault diagnosis.
Second, a new stochastic model is introduced to address this issue: it introduces a hidden variable to indicate the occurrence of the impacts and estimates the spectral content of the corresponding transients together with the spectrum of background noise. Results are found superior or at least equivalent to those of conventional envelope analysis and fast kurtogram.
Third, a novel scheme for extracting cyclostationary (CS) signals is proposed. It introduces a stochastic model, periodic-variance based model, to recover the CS component in the masking of interfering signals. Of particular interest is the robustness on experimental dataset and superior extraction capability over the conventional Wiener filter. Eventually, these experimental examples evidence its versatile usage for diagnostic analysis of compound signals.
Fourth, a benchmark analysis by using the fast computation of the spectral correlation is provided. This study benefits from two big data sets: one is the widely used open data set supplied by the Case Western Reserve University (CWRU) Bearing Data Center; the other is from the project of monitoring small fans in a production line measured by the author.

Informations complémentaires

  • Salle M1B (Bâtiment Saint Exupéry - INSA Lyon)

Mots clés